aboutsummaryrefslogtreecommitdiffstats
path: root/mm/memory_hotplug.c
diff options
context:
space:
mode:
authorTang Chen <tangchen@cn.fujitsu.com>2013-02-22 19:33:18 -0500
committerLinus Torvalds <torvalds@linux-foundation.org>2013-02-23 20:50:13 -0500
commita1e565aa3cfc7c6252cabc93de8391d12b9216aa (patch)
treef02974503acc8b61a7d8857c36dfb8b6cf7897ab /mm/memory_hotplug.c
parentd822b86a99e8d2b7ebbe3aba099288354287a885 (diff)
memory-hotplug: do not allocate pgdat if it was not freed when offline.
Since there is no way to guarentee the address of pgdat/zone is not on stack of any kernel threads or used by other kernel objects without reference counting or other symchronizing method, we cannot reset node_data and free pgdat when offlining a node. Just reset pgdat to 0 and reuse the memory when the node is online again. The problem is suggested by Kamezawa Hiroyuki. The idea is from Wen Congyang. NOTE: If we don't reset pgdat to 0, the WARN_ON in free_area_init_node() will be triggered. [akpm@linux-foundation.org: fix warning when CONFIG_NEED_MULTIPLE_NODES=n] [akpm@linux-foundation.org: fix the warning again again] Signed-off-by: Tang Chen <tangchen@cn.fujitsu.com> Reviewed-by: Wen Congyang <wency@cn.fujitsu.com> Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com> Cc: Jiang Liu <jiang.liu@huawei.com> Cc: Jianguo Wu <wujianguo@huawei.com> Cc: Kamezawa Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> Cc: Lai Jiangshan <laijs@cn.fujitsu.com> Cc: Wu Jianguo <wujianguo@huawei.com> Cc: Yasuaki Ishimatsu <isimatu.yasuaki@jp.fujitsu.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: "H. Peter Anvin" <hpa@zytor.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Diffstat (limited to 'mm/memory_hotplug.c')
-rw-r--r--mm/memory_hotplug.c24
1 files changed, 16 insertions, 8 deletions
diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
index 3409134ff20a..e189b1f4a9db 100644
--- a/mm/memory_hotplug.c
+++ b/mm/memory_hotplug.c
@@ -1017,11 +1017,14 @@ static pg_data_t __ref *hotadd_new_pgdat(int nid, u64 start)
1017 unsigned long zholes_size[MAX_NR_ZONES] = {0}; 1017 unsigned long zholes_size[MAX_NR_ZONES] = {0};
1018 unsigned long start_pfn = start >> PAGE_SHIFT; 1018 unsigned long start_pfn = start >> PAGE_SHIFT;
1019 1019
1020 pgdat = arch_alloc_nodedata(nid); 1020 pgdat = NODE_DATA(nid);
1021 if (!pgdat) 1021 if (!pgdat) {
1022 return NULL; 1022 pgdat = arch_alloc_nodedata(nid);
1023 if (!pgdat)
1024 return NULL;
1023 1025
1024 arch_refresh_nodedata(nid, pgdat); 1026 arch_refresh_nodedata(nid, pgdat);
1027 }
1025 1028
1026 /* we can use NODE_DATA(nid) from here */ 1029 /* we can use NODE_DATA(nid) from here */
1027 1030
@@ -1074,7 +1077,8 @@ out:
1074int __ref add_memory(int nid, u64 start, u64 size) 1077int __ref add_memory(int nid, u64 start, u64 size)
1075{ 1078{
1076 pg_data_t *pgdat = NULL; 1079 pg_data_t *pgdat = NULL;
1077 int new_pgdat = 0; 1080 bool new_pgdat;
1081 bool new_node;
1078 struct resource *res; 1082 struct resource *res;
1079 int ret; 1083 int ret;
1080 1084
@@ -1085,12 +1089,16 @@ int __ref add_memory(int nid, u64 start, u64 size)
1085 if (!res) 1089 if (!res)
1086 goto out; 1090 goto out;
1087 1091
1088 if (!node_online(nid)) { 1092 { /* Stupid hack to suppress address-never-null warning */
1093 void *p = NODE_DATA(nid);
1094 new_pgdat = !p;
1095 }
1096 new_node = !node_online(nid);
1097 if (new_node) {
1089 pgdat = hotadd_new_pgdat(nid, start); 1098 pgdat = hotadd_new_pgdat(nid, start);
1090 ret = -ENOMEM; 1099 ret = -ENOMEM;
1091 if (!pgdat) 1100 if (!pgdat)
1092 goto error; 1101 goto error;
1093 new_pgdat = 1;
1094 } 1102 }
1095 1103
1096 /* call arch's memory hotadd */ 1104 /* call arch's memory hotadd */
@@ -1102,7 +1110,7 @@ int __ref add_memory(int nid, u64 start, u64 size)
1102 /* we online node here. we can't roll back from here. */ 1110 /* we online node here. we can't roll back from here. */
1103 node_set_online(nid); 1111 node_set_online(nid);
1104 1112
1105 if (new_pgdat) { 1113 if (new_node) {
1106 ret = register_one_node(nid); 1114 ret = register_one_node(nid);
1107 /* 1115 /*
1108 * If sysfs file of new node can't create, cpu on the node 1116 * If sysfs file of new node can't create, cpu on the node